Osteochondral Tissue Formation Through Adipose-DerivedStromal Cell Differentiation on Biomimetic Polycaprolactone

Nanofibrous Scaffolds with Graded Insulinand Beta-Glycerophosphate Concentrations

Cevat Erisken, Ph.D.,1,* Dilhan M. Kalyon, Ph.D.,1,2 Hongjun Wang, Ph.D.,2

Ceren Ornek-Ballanco, Ph.D.,2 and Jiahua Xu, Ph.D.2

The ability to fabricate tissue engineering scaffolds containing systematic gradients in the distributions ofstimulators provides additional means for the mimicking of the important gradients observed in native tissues.Here the concentration distributions of two bioactive agents were varied concomitantly for the first time (oneincreasing, whereas the other decreasing monotonically) in between the two sides of a nanofibrous scaffold.This was achieved via the application of a new processing method, that is, the twin-screw extrusion andelectrospinning method, to generate gradients of insulin, a stimulator of chondrogenic differentiation, and b-glycerophosphate (b-GP), for mineralization. The graded poly(e-caprolactone) mesh was seeded with humanadipose-derived stromal cells and cultured over 8 weeks. The resulting tissue constructs were analyzed for andrevealed indications of selective differentiation of human adipose-derived stromal cells toward chondrogeniclineage and mineralization as functions of position as a result of the corresponding concentrations of insulin andb-GP. Chondrogenic differentiation of the stem cells increased at insulin-rich locations and mineralizationincreased at b-GP-rich locations.

Introduction

Native tissues are heterogeneous in structure andfunction that allow them to perform their physiological

requirements.1 Transitional interfaces between tissues playespecially crucial roles. For example, at the cartilage-to-bone(osteochondral) interface, the extracellular matrix (ECM)components, including cell types,2 as well as the concentra-tion of mineral particles3,4 vary as a function of distance tomeet the requirements of the complex hierarchical structureof native tissue. Various degenerative diseases, for example,osteoarthritis, deteriorate this structure, and the existingcartilage grafting methods have, so far, not been able toachieve stable integration with the cartilage or subchondralbone,5 leaving the regeneration of the native osteochondralinterface as a significant clinical challenge. This clinicalchallenge can be addressed using tissue engineering methodsprovided that the tissue engineering scaffolds can be tailoredto promote the mimicking of the gradients found at inter-faces, including cellularity and mineral content. One way ofcreating such gradients involves the use of functionally

graded scaffolds that are incorporated with bioactive agentsto modulate cell behavior selectively toward desired paths.However, the embedment of bioactive agents into tissueengineering scaffolds in a controlled manner is a challengerequiring specialized tools and processing techniques.6,7

Upon the recognition of the benefits of the utilization ofsignaling molecules in tissue engineering, various types ofgrowth factors have been exogenously added into cellgrowth media to direct cell behavior. For example, Luo et al.8

supplemented cell culture media with transforming growthfactor-beta 1, basic fibroblast growth factor, or hepatocytegrowth factor to examine their effects on the proliferationand ECM synthesis behavior of human vocal fold fibroblastsas well as on matrix contraction. In addition, the use of stemcells, which could be differentiated using appropriate me-diators, has opened up new avenues in tissue engineeringapplications.9–12 Growth factors, including recombinant hu-man insulin-like growth factor 1 and bone morphogeneticprotein 2, have been incorporated into scaffolds to inducestem cell differentiation into osteogenic and chondrogenicphenotypes through controlled release.10,11 Differentiation of

Departments of 1Chemical Engineering and Material Science and 2Chemistry, Chemical Biology and Biomedical Engineering, StevensInstitute of Technology, Hoboken, New Jersey.

*Current affiliation: Department of Biomedical Engineering, Columbia University, New York, New York.

TISSUE ENGINEERING: Part AVolume 17, Numbers 9 and 10, 2011ª Mary Ann Liebert, Inc.DOI: 10.1089/ten.tea.2009.0693

1239

human mesenchymal stem cells into osteogenic and chon-drogenic lineages using microencapsulated recombinanthuman bone morphogenetic protein 2 and recombinant hu-man insulin-like growth factor 1 with controlled doses alongthe length of the scaffold has also been demonstrated.13 A re-cent study of Benoit et al.14 has indicated that small-moleculefunctional groups attached to three-dimensional (3D) sub-strates can induce selective differentiation of mesenchymalstem cells, suggesting that tissue engineering scaffolds havethe potential to induce location-dependent changes in dif-ferentiation of stem cells, provided that scaffolds gradedwith the requisite chemical composition, pore size, and po-rosity distributions are available. Overall, these challengesare compounded by the urgent need for the translation oflaboratory-based and generally labor-intensive, multi-stepscaffold preparation techniques into ‘‘clinically appropriatelarger scale production techniques’’ that are ‘‘reproducible,safe, clinically effective, and economically acceptable.’’15

In this study, nanoparticles of insulin and b-glyceropho-sphate (b-GP) were incorporated into a nonwoven mat ofpoly(e-caprolactone) (PCL) nanofibers in a graded manner,that is, controlled concentration distributions as a function ofscaffold thickness. Human adipose-derived stromal cells (h-ADSCs) were cultured on these nonwoven graded mats forprobing their effects on the development of cellularity andmineralization. It is hypothesized that the encapsulation ofstimulators into biodegradable nanofibers in a spatially con-trolled manner would guide the selective differentiation ofstem cells into chondrocytes and relevant ECM deposition.More specifically, gradients in insulin concentration along thethickness of the scaffolds would induce gradients of numberof chondrocytes as well as chondrogenic matrix deposition,the gradients being in line with insulin concentration.

Similarly, the gradients in concentration of b-GP wouldlead to incremental mineralization in the tissue constructs.The structure generated in tissue constructs could then allowfor better mimicking of the biological and physical propertiesof native osteochondral ECM. The aim of this study arethreefold: (i) to create concentration gradients of multipleingredients, that is, insulin and b-GP, in nanofibrous scaf-folds, (ii) to show the functionality of the scaffolds to selec-tively differentiate the h-ADSCs into chondrocytes in agraded manner as affected by insulin concentration gradient,and (iii) to generate physical mineralization gradients as af-fected by b-GP. For the fabrication of the graded scaffolds,a recently developed hybrid twin-screw extrusion and elec-trospinning technology was utilized.16 This method enablesthe fabrication of tissue engineering scaffolds with gradientsin composition, porosity, wettability, and mechanical prop-erties.17 This technology is also suitable to translate thelaboratory-based and generally labor-intensive, multi-stepscaffold preparation techniques into clinically appropriateand economically feasible production scales.

The critical attributes of the materials of the study in-cluded the biocompatibility, ease of processability, and re-lease capability of PCL, as well as the abundant availability(from lipoaspirate) and the demonstrated multi-lineage po-tential (including chondrocytic and osteoblastic lineages18) ofthe h-ADSCs. The use of insulin relied on its demonstratedability for chondrogenic differentiation of mesenchymalcells.19,20 Insulin has also been widely used as a supplementto stimulate the differentiation of adipose-derived stem

cells,18 human bone marrow cells,21 and cartilage progenitorcells22 into chondrogenic phenotypes. The scaffolds werealso incorporated with b-glycerophosphate (b-GP) to pro-mote mineralization,23 on the basis of elevated local phos-phate ion concentrations induced by alkaline phosphatase ormodulation of osteoblast function by phosphate ions throughincreased bone remodeling or collagen synthesis.24,25

Materials and Methods

Scaffold fabrication

Continuously graded insulin/PCL/b-GP scaffolds werefabricated using the recently developed hybrid twin-screwextrusion/electrospinning technology.16 The process consistsof a twin-screw extruder with fully intermeshing and co-rotating screws (Material Processing & Research, Inc.) inte-grated with a spinneret die with multichannels for flow andshaping, and connected to a high-voltage supply for elec-trospinning. The tight clearances between screws and thebarrel, and between the screws themselves enabled thebreak-up of agglomerates while keeping additives distrib-uted homogeneously in the suspension. Operation of twin-screw extruder was controlled by the aid of a computerizedfield-point based data acquisition/process control system.Electrospinning of nanofibers occurred at a potential of 5 kVwith a distance of separation of 7.5 cm between the collectingplate and spinneret die, using a spinneret with a die diameterof 0.6 mm. The ability to tightly control the feed rates andoperating conditions provides the wherewithal necessary tofabricate tissue engineering scaffolds continuously and re-producibly (versus batch type cyclic processes).

The PCL (Sigma-Aldrich; Catalog# 440744) was dissolved indichloromethane (DCM) at a ratio of 12:100 (g/mL), and in-sulin (Sigma-Aldrich; Catalog# I1882) and b-GP disodium salthydrate (b-GP; Sigma-Aldrich; Catalog# G9891) were added toPCL/DCM solution to obtain ultimate insulin/PCL and b-GP/PCL concentration ratios of 25:75 and 50:50 by weight, re-spectively. The b-GP was pulverized before use to have a finalparticle diameter in the range between 100 and 900 nm. Insulinwas received in the form of a nanosized powder and was usedas procured (Fig. 1c). The insulin/PCL/DCM suspension wasfed into the first mixing zone of the extruder into which b-GP/PCL/DCM was also introduced in a time-dependent, that is,continuously increasing, manner, whereas the feed rate of in-sulin/PCL/DCM was concomitantly decreased. The nonwo-ven meshes (thickness of *500mm) with gradients ofconcentrations of insulin and b-GP were, then, punched intodiscs of 8 mm in diameter to be used as scaffolds for culturingof h-ADSCs or for mechanical characterization.

Tissue formation

Discs were first sterilized in 70% ethanol and then washedby immersing in a phosphate-buffered saline (PBS). The h-ADSCs were obtained as part of a series of adipose stromalcells isolated from 13 patients, which were separately testedby Coriell Institute for Medical Research, for their osteogenic,chondrogenic, and adipogenic differentiation. It was deter-mined that 12 of 13 isolations showed similar capacity forosteogenic, chondrogenic, and adipogenic differentiationupon culture in the corresponding differentiation media(private communication). One out of these 12 isolations

1240 ERISKEN ET AL.

(obtained from the neck and arms of a 56-year-old womanupon tumescent liposuction from subcutaneous adipose tis-sue) was used in our study. Cells were proliferated in a lowserum (0.5% fetal bovine serum) medium supplementedwith epidermal growth factor, platelet-derived growth factorBB, insulin, and transferrin. Before seeding, cells (at passage5) were detached with 0.25% trypsin–EDTA (Invitrogen;Catalog# R-001-100), centrifuged, and resuspended in com-bined medium (described below) for homogeneity.

A total of 34,000 cells17 were seeded onto each side of thescaffolds placed in a 24-well plate. Then, 1 mL of combinedmedium was added to immerse the scaffolds completely,followed by their culturing in an incubator at 378C and 5%CO2. The combined medium was prepared by mixingMinimum Essential Medium (MEM) Alpha Medium (In-vitrogen; Catalog# 11900-073), which was supplementedwith fetal bovine serum (Invitrogen; Catalog# 16000-036),penicillin/streptomycin (1/0.045/0.01 volume ratio, respec-tively), and NaHCO3 (2.2 g/L) with Dulbecco’s modifiedEagle’s medium high glucose (Invitrogen; Catalog# 11960-051) supplemented with fetal bovine serum and penicillin/streptomycin (1/0.1/0.01 volume ratio, respectively) by aratio of 50:50 by volume. MEM Alpha and Dulbecco’smodified Eagle’s medium high glucose were successfullyused earlier for osteoblastic26 and chondrogenic27 differen-tiation, respectively, in conjunction with various types ofstem cells, including human adipose-derived stem cells. Themedium was replaced every other day, and cell–scaffoldconstructs were flipped upon every change of the medium.Such flipping of scaffolds was carried out to reduce thesedimentation of the cells. Resulting tissue constructs wereharvested after 1, 4, and 8 weeks and were kept in 4%buffered formaldehyde solution until characterization.

Scaffold characterization

A LEO Gemini 982 scanning electron microscope wasused for the characterization of the nanofiber diameters, poresize, and the surface attributes of electrospun meshes. Thepresence of b-GP was assessed via Na and P scans at 10 kV.The spatial distributions of the concentrations of insulin andb-GP in the scaffolds were measured by using a thermogravimetric analysis, TGA, apparatus (TGA-Q50, TA In-struments). The TGA experiments were carried out byheating of the specimens collected from different locationsin the scaffolds from 258C to 5908C at a constant rate of158C/min under N2 atmosphere.

Tissue processing and characterization

Fixed tissue constructs were dehydrated using a series ofincreased concentrations of ethanol and kept overnight undervacuum. For scanning electron microscopy (SEM) character-ization, the specimens were mounted on Al stubs, coated withAu, and analyzed using a LEO Gemini 982 scanning electronmicroscope. For histological analyses, the harvested tissueconstructs were embedded in a glycol methacrylate-basedembedding material, that is, Immuno-Bed (Polyscience, Inc.).Thin sections (5mm) were obtained from tissue constructsembedded in plastic and stained with hematoxylin and eosin(H&E) to examine the cell distribution and chondrocyte-specific cell morphology. Glycosaminoglycan (GAG) andcollagen depositions by chondrocytes were evaluated after 4weeks using Alcian Blue (AB) and picrosirius red stains, re-spectively. The intensity of the stain over the entire tissueconstructs was quantified using ImageJ software. To furtherprobe the differentiation of h-ADSCs into chondrogenic phe-notype, tissue constructs after 1 and 8 weeks of cell culture

FIG. 1. Scanning electronmicroscopy micrographs ofgeneral surface structure offunctionally graded scaffolds(a), with higher magnifica-tions of b-GP-rich (b), andinsulin-rich (c) sides. b-GP,b-glycerophosphate.

BIOMIMETIC SCAFFOLDS WITH GRADED MULTIPLE STIMULATORS 1241

were permeabilized and incubated with a goat polyclonalanti-melanoma-inhibiting activity (MIA), a rabbit polyclonalanti-activin receptor-like kinase (ALK)-1, and anti-fibroblastsgrowth factor receptor 3 (FGFR3), and subsequently FITC-conjugated secondary antibodies. Samples were imaged usinga Zeiss LSM510 confocal microscope. Z-stack feature was usedto detect the spatial distribution of the chondrocytic differ-entiation within the tissue constructs. Sections were alsotreated with AgNO3 solution and subjected to strong light(i.e., von Kossa stain) to evaluate the presence of Ca. Thestained sections were analyzed using a Nikon Polarizing Mi-croscope (Micron Optics). The distribution of mineral depo-sition over the entire scaffolds and time were evaluated byquantifying the stain intensity using ImageJ software.

Native cartilage-bone specimen preparation

Fresh bovine knee joints (6 months old) were obtainedfrom a local abattoir and stored at �808C. The joints werethawed at room temperature for *12 h and femoral condyleswere drilled to a depth of *3 mm to remove the native os-teochondral tissue. The specimens were equilibrated in PBSsolution at 48C. Thicknesses were measured at four differentlocations and samples with 100 mm or more variations werediscarded.

Biomechanical characterization

Unconfined uniaxial compression testing of engineeredtissue constructs and the native osteochondral tissue wereperformed using a Rheometric Scientific ARES Rheometer(currently TA Instruments) in the strain amplitude range of0% to 10%. The specimens while in their fresh state were firstequilibrated under a tare load and then compressed at aconstant rate of 0.05 mm/s. The modulus and toughnessvalues were determined from the slope of stress versus strainbehavior and the area under the stress versus strain curve,respectively. To prevent dehydration during tests, specimenswere deformed within a custom-designed environmentalchamber28 filled with PBS and kept at 378C.

Statistical analysis

All statistical analyses were performed using SPSS 13.0statistical software. Assumptions of normality and homoge-neity of variance were verified by using Shapiro-Wilks’s testand Levene’s test, respectively, and homogeneity of groupcovariance matrices was verified by using Box’s M test.Statistical significance for the change of insulin and b-GPconcentrations as well as mineral distributions along thethickness of scaffolds was investigated using one-way mul-tivariate analysis of variance (ANOVA). Significant multi-variate omnibus test statistics were followed by univariateanalyses. Type I error was controlled by employing theBonferroni correction and testing each ANOVA at a signifi-cance (alpha) level of 0.025 (i.e., p¼ 0.05/2). The Bonferronipost-hoc procedure was then used to test for significance incases of multiple comparisons. The distributions of chon-drocytic cell count, GAG, and collagen deposition along thelength of tissue constructs were examined through one-wayANOVA, followed by Tukey’s post hoc comparisons. Themineral deposition over time was evaluated using Student’st-test.

Results

Scaffold fabrication and characterization

Scaffolds consisting of highly porous nonwoven mesheswith pore sizes ranging between 5 and 50mm and fiber di-ameters in the range of 200 to 2000 nm were fabricated and areshown in Figure 1a. Bottom, b-GP-rich, and top, insulin-rich,sides of the typical scaffolds are also shown in Figure 1b and c,respectively. Location-dependent thermo-gravimetric charac-terization of the scaffold meshes validated that indeed con-tinuous concentration gradients of both insulin and b-GPcould be achieved along the thickness of the scaffolds (i.e.,along 500mm). The concentration of insulin increased linearlyfrom 0% to 23.2%� 2.6% by weight from one side of thescaffold to the other, whereas the b-GP concentration de-creased from 48.3%� 2.8% to 0% by weight along the samedirection (Fig. 2a). The concentration (weight fraction) of b-GPat the bottom of the scaffold (48.3%� 2.8%, n¼ 3) is signifi-cantly higher than that of b-GP at 175mm (14.1%� 0.7%, n¼ 3;p¼ 0.000), and that of b-GP at 300mm (1.0%� 1.0%, n¼ 3;p¼ 0.000). Similarly, the concentration of insulin at the top ofthe scaffold (23.2%� 2.6%, n¼ 3) is significantly higher thanthat of insulin at 300mm (15.8%� 3.1%, n¼ 3; p¼ 0.002), andat 175mm (7.8%� 0.5%, n¼ 3, p¼ 0.000). Energy dispersivespectra provided collaborative data, that is, decreasing inten-sities of Na and P peaks, characteristic of b-GP, with in-creasing distance from the bottom of electrospun mesh,indicative of the decreasing concentration of b-GP (Fig. 2b).

PCL scaffolds containing either only b-GP or only insulinwere also fabricated and immersed in cell culture media toevaluate the time-dependent release of b-GP and insulin. Therelease study was carried out to determine the concentrationsof bioactive agents to be incorporated into the PCL scaffolds.The weight loss of the scaffolds containing 23% wt insulin is

FIG. 2. Thermo gravimetric analysis, TGA (a) and energydispersive spectroscopy, EDS (b) characterization for theconcentration distribution of bioactive agents. Symbols * and** indicate significance at p< 0.025 for insulin and b-GP, re-spectively, for n¼ 3. Error bars represent standard error(s.e.m.) at 97.5% confidence level. Color images availableonline at www.liebertonline.com/tea

1242 ERISKEN ET AL.

5% in two days (Figure 3c), which corresponds to the releaseof about 22% of the insulin initially loaded, assuming that thePCL is not affected. There is negligible release of the insulinafter two days. It was observed that b-GP, which is a hy-drophilic Na salt, released at a higher rate in the media due toits greater dissolution rate.29 After 6 weeks, the surface mor-phology of scaffolds was observed to change from smooth toirregular and also to exhibit surface voids (Fig. 3). Surfaceporosity was more prominent on the b-GP-rich side (Fig. 3a),associated with greater dissolution rate (Fig. 3c) of the hy-drophilic Na salt, that is, b-GP, in comparison to insulin.

Tissue formation and SEM characterization

Attachment, proliferation, and differentiation behavior ofh-ADSCs were investigated during the 8 weeks of cultur-ing. SEM characterization of tissue constructs revealed thatcells attached well over the entire length of the scaffold(Fig. 4a, d), and cell proliferation took place with multi-ple cell layers forming at the sides (Fig. 4b, e) and inte-rior volume, that is, the ‘‘core,’’ of the scaffolds (Fig. 4c, f, g).The formation of collagen fibril bundles (inset to Fig. 4c)is apparent in insulin-rich side after 8 weeks of culture.

FIG. 3. Scanning electron microscopy micrographs of surface structure of b-GP-rich (a), and insulin-rich (b) sides after therelease of bioactive agents together with weight-based release profile for a duration of 6 weeks (c).

FIG. 4. Extracellular matrix formation and mineralization of tissue constructs. Cells adhered well to both sides of scaffoldsafter 1 week (a, d). Globular particulates (supposedly minerals) are apparent in b-GP-rich side (d), whereas no particulateswere observed in insulin-rich side (a). After 4 weeks, surfaces of scaffolds were covered with layers of cells (b, e) with greaterparticulate deposition in b-GP-rich side (e). After 8 weeks, formation of extracellular matrix (c, f) and fibrils (inset to c)revealed further development of tissue. Cross section of tissue constructs after 8 weeks (g) also indicated formation of richand highly mineralized extracellular matrix in b-GP-rich side.

BIOMIMETIC SCAFFOLDS WITH GRADED MULTIPLE STIMULATORS 1243

The formation of globular mineral particulates (Fig. 4d, e, g)in b-GP-rich regions suggested that b-GP was indeedeffective in promoting the greater extent of mineraliza-tion of tissue constructs at locations where it was readilyavailable.

Histology and immunohistochemistry

H&E staining was performed on tissue constructs after 1and 8 weeks of cell culture and the results are shown in Figure5a, b, and d. Cells with distinctly different morphologies in-

FIG. 5. Hematoxylin and eosin staining showed proliferation and differentiation of h-ADSCs after 1 week (a) and 8 weeks(b). Differentiation to chondrocyte-specific round morphology (shown with black arrows in d) from h-ADSCs (c, monolayer,stained with TRITC-conjugated phalloidin for filament-actin [Red/Orange] and DAPI for cell nucleus [blue]) is evident,although some h-ADSCs (white arrow in d) remained undifferentiated. Chondrogenic differentiation of h-ADSCs was alsodemonstrated by immunostaining for melanoma-inhibiting activity (MIA) (top: e, f; bottom: h, i). Confocal microscopyanalysis showed increasing intensity of MIA immunostaining on the top section of scaffold as a function of time (e, f). Twomorphologically distinct cell populations, that is, differentiated and dedifferentiated chondrocytes, were immunodetectedwith ALK-1 (g, j). Cells with predominantly round morphology are shown in (g). h-ADSC, human adipose-derived stromalcell; ALK, activin receptor-like kinase. Color images available online at www.liebertonline.com/tea

FIG. 6. Immunostaining for FGFR3 after 8 weeks. (a–c) Horizontal slices demonstrating depth-dependent change in FGFR3intensity over the scaffold (insulin rich to b-GP rich from left to right), and (d–f) vertical cross sections of the scaffold showingdepth-dependent variation in FGFR3 intensity (top: insulin rich, bottom: b-GP rich). FGFR3, fibroblasts growth factor re-ceptor 3. Color images available online at www.liebertonline.com/tea

1244 ERISKEN ET AL.

volving the characteristic round shape of the chondrocytesand the h-ADSCs (Fig. 5c) are observed. The chondrocyticdifferentiation of h-ADSCs was further confirmed with de-tection of two type II collagen-coexpressing chondrogenesismarkers, MIA30–33 and FGFR3.34–38 Confocal microscopyanalysis showed the degree of intensity of both markers on thetissue constructs as a function of insulin/b-GP gradient (Figs.5e, f, 6). Increasing intensity of MIA immunostaining wasobserved on the top section of scaffold (insulin-rich) in week 8constructs (Fig. 5e, f). Vertical and horizontal sections of week8 constructs stained with FGFR3 demonstrated the highestFGFR3 intensity on the insulin-rich sections (Fig. 6a–c), and aclear gradient of FGFR3 intensity proportional with the gra-dient of insulin concentration (Fig. 6d–f). A 3D construction ofthe FGFR3 intensity over the entire scaffold is also provided asSupplementary Movie S1 (Supplementary Data are availableonline at www.liebertonline.com/tea). As differentiatedchondrocytes are characteristically unstable by undergoingdedifferentiation and/or re-differentiation,39–43 phenotypicstability of chondrocytes was also evaluated by immuno-staining the tissue constructs (Fig. 5g, j) with ALK-1, a markerof dedifferentiated chondrocytes.34 ALK-1 was detected intwo morphologically distinctive cell populations, that is, dif-ferentiated and dedifferentiated chondrocytes, in the 8-weektissue constructs corresponding to the gradient of insulin andb-GP.

The number density of chondrogenic cells (with character-istic round shapes) increased with distance from the bottomsurface of tissue constructs, consistent with the monotonicincrease of insulin concentration over the same region (Fig.7a). The number density of these cells correlated positively(r2¼ 0.95) with concentration of insulin (Fig. 7b), suggestingthat the local availability of insulin was effective in promotingthe preferential differentiation of stromal cells. The countof chondrocyte-like cells showed that the number of chon-drocytes located at distances in the range of 330 to 500mmfrom bottom, that is, 39.64%� 2.54%, is significantly greaterthan those located at distances between 220 and 330mm fromthe bottom surface of the scaffold, that is, 24.23%� 0.85%(n1¼n2¼ 6, a¼ 0.05, p¼ 0.000). Similarly, the number ofchondrocytes located at distances between 220 and 330mmfrom bottom were found to be significantly greater than thoselocated at distances between 110 and 220mm, that is,19.46%� 1.73% (n1¼n2¼ 6, a¼ 0.05, p¼ 0.006).

The AB stains performed on tissue constructs after 4weeks of culturing (Fig. 8a) suggested greater (yet statisti-cally insignificant, p> 0.05) staining intensity for sulfatedGAG with increasing insulin concentration (Fig. 8b). On theother hand, the picrosirius red stain (Fig. 8c) revealed sta-tistically significant effect of increasing insulin concentrationon increasing total collagen deposition (Fig. 8d). The vonKossa stains performed on tissue constructs after 1 week(Fig. 9a) and 8 weeks (Fig. 9b) of culturing showed positivestaining for Ca3(PO4)2. b-GP incorporated into the nanofibersof our scaffolds generally gave rise to an apparent trend ofgreater mineral deposition at b-GP-rich locations (Fig. 9c)and over time (Fig. 9d).

Biomechanical characterization

The comparisons of compressive properties of tissue con-structs obtained at different culture durations versus those of

native bovine osteochondral tissue are shown in Figure 10.Engineered tissue constructs and the native tissue both ex-hibited linear behavior in 0% to 10% strain amplitude rangeupon compression at 0.05 mm/s. Compression, at 10% strain,of tissue constructs obtained after 1, 4, and 8 weeks yieldednormal stress values of 0.674� 0.06, 0.852� 0.121, and1.916� 0.100 kPa, respectively. Significantly higher modulus(slope of stress versus strain curve) and toughness (areaunder stress versus strain curve) values were obtained after8 weeks of cell culture as compared to those obtained after1 and 4 weeks, suggesting a considerable degree of tissueformation within 8 weeks (Fig. 10b, c). The modulus valuesof the native tissue, that is, 20.9� 0.15 kPa, were greater thanthose of engineered tissue constructs at 8 weeks, that is,0.19� 0.01 kPa. Similarly, the modulus values of engineeredtissue constructs at 8 weeks were greater than those of en-gineered tissue constructs at 4 weeks, that is, 0.09� 0.01 kPa.As for toughness, the values of native tissue, that is,1023� 33 kPa, were greater than those of engineered tissueconstructs at 8 weeks, that is, 9.64� 0.70 kPa. Similarly,toughness values of engineered tissue constructs at 8 weeks

FIG. 7. Count of chondrocyte-like cells (a) showed that thenumber of chondrocytes increased from the bottom of thescaffolds to the top. Symbol * indicates significance at p< 0.05for n¼ 6. Error bars represent s.e.m. at 95% confidence level.There was appreciable correlation (r2¼ 0.95) between cellfrequency and insulin concentration along the scaffoldthickness (b).

BIOMIMETIC SCAFFOLDS WITH GRADED MULTIPLE STIMULATORS 1245

were determined to be greater than those of engineered tis-sue constructs at 4 weeks, that is, 4.53� 0.41 kPa.

Discussion

Aiming at hierarchical structures of interfaces, scaffolds thatwere spatially organized in the form of distinct multi-lay-ers5,23,44–48 have been fabricated and culturing of single44 ormultiple5,46 cell types on these multi-layered scaffolds havebeen studied. Schaefer et al.,46 for example, prepared two in-dividual sheets consisting of polyglycolic acid on one side anda blend of poly-lactic-co-glycolic acid (PLGA) and polyethyl-ene glycol on the other, and then seeded the two sides with

chondrocytes and periosteal cells, respectively. The two con-structs were then sutured together after 1 week of cultureperiod to obtain 3D cartilage–bone composites. Later, Sher-wood et al.47 designed and manufactured cartilage/bonescaffolds by layering three different materials composed ofD,L-PLGA/L-poly(lactic acid) (PLA) for cartilage, L-PLGA/tricalcium phosphate (TCP) for bone, and a blend of thesematerials for the bone–cartilage transition zone. In anotherstudy, Allan et al.23 developed biphasic constructs consistingof cartilaginous tissue anchored to the top surface of a bonesubstitute with a calcified interface. These constructs wereseeded with deep zone chondrocytes from bovine articularcartilage, and cultured in the presence of b-glycerophosphate.

FIG. 8. Alcian Blue stain (a)and picrosirius red stain (c)after 4 weeks revealed gly-cosaminoglycan (b) and in-creasing collagen (d)deposition as a function ofincreasing insulin concentra-tion from bottom to top.Symbol * indicates signifi-cance between consecutivelocations. Error bars represents.e.m. at 95% confidencelevel (n¼ 3). Color imagesavailable online at www.liebertonline.com/tea

FIG. 9. von Kossa stain after1 week (a) and 8 weeks (b) ofculture, and mineral deposi-tion as determined by thestained area of sections as afunction of scaffold thickness(c) and time (d). Error barsrepresent s.e.m. at 97.5% and95% confidence level for (c)and (d), respectively, (n¼ 3).Symbol * indicates signifi-cance at p< 0.05.

1246 ERISKEN ET AL.

The formation of calcified cartilage region adjacent to thesubstrate and a hyaline-like zone above was observed.

More recently, Ng et al.48 formed bilayered scaffolds withdepth-varying compressive material properties and seededzonal populations of chondrocytes. They observed that suchbilayered constructs led to depth-dependent cellular andcompressive mechanical inhomogeneity, similar to that ofthe native tissue. A microparticle-based scaffold fabricationtechnique was also introduced recently to create scaffoldswith spatial control over active ingredients using uniformpoly(D,L-lactide-co-glycolide) microspheres with the aim ofdemonstrating the usefulness of such macroscopic gradientsfor interfacial tissue regeneration.49

In this study, biomimetic scaffolds made of unitary non-woven meshes that were incorporated with bioactive agents

were fabricated and employed to probe their suitability forthe selective differentiation of stem cells for osteochondralapplications. The highly porous scaffolds consisted of non-woven meshes with pore sizes ranging between 5 and 50mm.This pore size range is considered to be appropriate for cellinfiltration during culture.50 The ability to generate contin-uous concentration gradients of bioactive agents is an im-provement over the earlier studies in functional grading offibrous tissue engineering scaffolds, which typically utilizedonly one bioactive, which could be applied only in verysimplistic modes. For example, varying concentrations ofonly b-tricalcium phosphate from the top to bottom volumesof a nanofibrous mesh were previously achieved.17 Similarly,Li et al.51 demonstrated the grading of calcium phos-phate along the length of a nonwoven mat of electrospun

FIG. 10. Biomechanical properties of engineered tissue constructs and native osteochondral tissue under unconfinedcompression. Normal stress versus strain behavior under 10% compression (a), and corresponding modulus (b) andtoughness (c) values. Symbol * indicates significance at p< 0.05 for n¼ 3. Error bars represent s.e.m. at 95% confidence level.Symbols in (a): ^, tissue constructs at 1 week; *, at 4 weeks; &, at 8 weeks; D, the native tissue.

BIOMIMETIC SCAFFOLDS WITH GRADED MULTIPLE STIMULATORS 1247

nanofibers, and Phillips et al.3 generated gradients of retro-virus encoding the osteogenic transcription factor Runx2/Cbfa1 in collagen scaffolds containing poly(L-lysine). How-ever, to our knowledge there are no reports of the incorpo-ration of two or more bioactives into nanofibrous scaffolds ina continuously graded manner.

The ability of incorporating multiple bioactive agentswithin a nanostructured mat, that is demonstrated here,should provide a greater degree of freedom in mimickingboth the physical as well as biological structures of os-teochondral interface. The monotonic increase of b-GP andthe monotonic decrease of insulin from one side of thescaffold to the other is but one of the many types of grada-tions of the bioactive agents that are possible with the twin-screw extrusion and electrospinning process. This particulartype of gradation was selected to better isolate the individualeffects of the two bioactive agents as well as the effect of theircombination to mineralization and the differentiation ofh-ADSCs.

The rapid dissolution rate of b-GP is consistent with therapid achievement of the elevated concentrations of Ca andphosphates that are required for the initiation of minerali-zation.52 Insulin, on the other hand, is an amphiphilicpeptide; that is, it has a hydrophobic core consisting of car-bon-rich amino acids at its center and relatively hydrophilicouter surfaces consisting of charged amino acids.53 Thisamphiphilic nature of insulin gives rise to reduced dissolu-tion and diffusion rates. In the scaffolds, insulin concentra-tion increased linearly from 0% to 23.2%� 2.6% by weightfrom one side of scaffold to the other, whereas the b-GPconcentration decreased from 48.3%� 2.8% to 0% by weightalong the same direction. Based on the time-dependentweight loss of PCL-only, insulin/PCL, and b-GP/PCL me-shes after immersing into the media, the 23.2%� 2.6% insulinconcentration in the mesh was calculated to supply *25mg/mL of insulin to cell culture media during the first 2 days ofculturing. This concentration of insulin is appropriate be-cause it is within the concentration range at which effectivechondrogenic differentiation could be achieved.19,20,54 Con-centrations of insulin <40 mg/mL were determined to pro-mote chondrogenesis, whereas larger quantities (>40mg/mL) were found to downregulate the rate of chondrogen-esis.19 The sensitivity of chondrogenesis to insulin concen-tration was determined to be principally due to the effect ofinsulin to reduce the number of cell membrane receptorsprogressively by forming internalized hormone–receptorcomplexes at relatively high concentrations.54 Relatively highconcentrations of b-GP were targeted because elevated localphosphate ion concentrations are required to induce miner-alization.24,25

H&E staining results shown in Figure 5 indicated that thecells penetrated into the core of the scaffold and that thetrend of increasing number density of chondrocytic cells (Fig.7) (cells with characteristic round shapes55) is in line with themonotonic increase of insulin concentration, affected by thedifferences in local insulin concentration. This observation isconsistent with earlier findings,19,56,57 which have demon-strated the effectiveness of insulin for the differentiation ofstem cells. The qualitative nature of the distribution of thechondrocytic cell numbers is similar to what is observed tooccur at the native cartilage–bone interface. McLauchlan andGardner characterized human articular cartilage as a func-

tion of thickness and found that the density of chondrocytesdeclined from the superficial zone to the bone.2

Sulfated GAG was characterized and quantifiedthroughout the thickness of tissue constructs using AB dye.58

Since GAG is present in articular cartilage as a side chain ofproteoglycans, which are important determinants of cartilagemechanical properties,59 deposition of GAG (Fig. 8a) withinour engineered tissue constructs indicates the developmentof a cartilage-like structure. However, there was no statisti-cally significant dependence of GAG ( p> 0.05) on the localinsulin concentration (Fig. 8b). Collagen type II and type Iare widely utilized to mark the chondrogenic60–63 and oste-ogenic64–66 phenotypes, respectively. Collagen is the mostabundant macromolecule component (about 60% of the dryweight) in the articular cartilage. Particularly, the type IIcollagen represents 90% to 95% of the total collagen in car-tilage ECM, whereas other types (I, IV, V, VI, IX, and XI)constitute only a minor proportion.67 With regard to bonetissue, the major collagen type is found to be predominantlytype I.68,69 To probe the effects of the insulin gradient further,the tissue constructs of this study were stained with picro-sirius red stain,70 which binds to the collagen types I, II, andIII (Fig. 8c). The semi-quantitative analysis of the picrosiriusred stain results (Fig. 8d) revealed a statistically significantchange in the stain intensity with scaffold thickness. The totalcollagen indeed increased with increasing distance towardthe insulin-rich side. Thus, picrosirius red staining revealedsignificant increases in collagen synthesis with increasinginsulin concentrations following the insulin gradient thatwas built into the scaffold (Fig. 8d), on the basis of chon-drogenic differentiation as well as stimulative effect of insu-lin. Chondrogenically differentiated ADSCs were previouslyreported to produce more GAG and collagen than ADSCscontrols,71–73 suggesting that the graded collagen andGAG contents found in the tissue constructs of this study areattributable to the gradients of chondrogenically differenti-ated cells occurring as a result of the graded insulin con-centration. This is actually consistent with the study ofMusselmann et al., who have shown that insulin alonestimulates the collagen synthesis in vitro.74 Concomitant ex-amination of the H&E staining results of Figure 5 and thetotal collagen distribution given in Figure 8c and d suggestthat h-ADSCs penetrated across the entire cross section of thescaffold and that the distribution of the collagen depositiondeveloped as a function of the insulin concentration distri-bution.

The differentiation of h-ADSCs was also confirmed byimmunostaining of tissue constructs for MIA, that is, acartilage-derived retinoic acid-sensitive protein that has beenused as a marker for chondrocytic differentiation.33 Underphysiological condition, MIA is expressed only during thecourse of chondrogenesis through development and inmature chondrocytes and is not observed in other non-cartilaginous tissues.30,31,75 Although MIA is pathologicallyexpressed in malignant melanoma, minimal expression wasdetected in normal melanocytes.76 In addition, it was foundthat MIA-deficient mice developed structural abnormalitiesin cartilage.77 MIA is co-regulated with type II collagen inboth expression in vivo and tissue distribution.30–32 Similarly,differentiated chondrocytes in cell culture co-express MIAand collagen II.33 In addition to examining MIA, we alsodetected increasing intensity of FGFR3 with increasing in-

1248 ERISKEN ET AL.

sulin concentration (Fig. 6). FGFR3 is one of molecularmarkers during chondrogenesis and chondrocyte differenti-ation.38 FGFR3-null mice are defective in articular cartilagewith early arthritis.78 Further, its expression correlates withcollagen II by differentiated chondrocytes.34 The observationof increasing FGFR3 intensity with increasing insulin con-centration thus provided additional validation of the greaterrate of differentiation of h-ADSCs into chondrogenic phe-notype with increasing insulin concentration.

Differentiated chondrocytes are characteristically unstableby undergoing dedifferentiation and/or re-differentiation.39–43

Phenotypic stability of chondrocytes was also evaluated byimmunostaining the tissue constructs with ALK-1, a receptorinvolved in hypertrophic maturation of chondrocytes79 and amarker of dedifferentiated chondrocytes.34 Thus, we alsostained ALK-1 to evaluate the phenotypic stability of thedifferentiated chondrocytes. Interestingly, ALK-1 was im-munodetected in two morphologically distinctive cell popu-lations in the 8-week tissue constructs corresponding to thegradient of insulin and b-GP. These results indicate that par-tial dedifferentiation occurred after chondrogenesis; however,a large portion of stable chondrogenic population remainedin tissue constructs after 8 weeks in culture.

Regarding the effect of b-GP, the scaffold regions thatwere richer in the concentration of b-GP gave rise to greatermineral deposition at the same location as revealed by vonKossa staining. This finding is in agreement with the resultsof Allan et al., who utilized b-GP for the mineralization ofbiphasic constructs comprised of cartilaginous tissue an-chored to the top surface of a bone substrate.23 The exactmechanism by which b-GP induces mineralization is un-clear.52 Generally, it is accepted that organic phosphate ishydrolyzed by alkaline phosphatase, an enzyme found oncell membrane of bone cells, to release free inorganic phos-phate (iP), thus providing a chemical potential for mineraldeposition. The iP is believed to induce mineralization eitherby direct mineral deposition due to elevated local iP con-centration24,80 or by metabolic process involving protein andRNA synthesis.52,81–83

In the characterization of biomechanical properties ofcartilage tissue as well as engineered tissue constructs, ingeneral, the aggregate modulus and the permeability arecharacterized as the parameters of well-studied cartilage bi-phasic model.84 Here the unconfined compression technique,that is, similar to the characterization of tensile properties ofthe constructs in the radial direction along with the com-pressive properties in the thickness direction,85 was usedprimarily on the basis of analogies between physiologicalcontact loading and unconfined compression.86 Elasticmodulus (slope of stress versus strain curve in the linearrange) and area under stress versus strain curve of tissueconstructs were previously reported to be indicators of ex-tent of mineralization and collagen formation, respective-ly.87,88 The significantly higher modulus and area understress versus strain curve values obtained after 8 weeks ofcell culture as compared to those obtained after 1 and 4weeks, therefore, suggest a considerable degree of tissueformation within 8 weeks.

In summary, the recently developed hybrid extrusion/electrospinning process was successfully implemented forgenerating tissue engineering scaffolds with controlled gra-dations of concentrations of insulin and b-GP. To our

knowledge, this is the first time that multiple bioactive in-gredients were incorporated in a distributed manner innanofibrous-type biomimetic scaffolds. In this demonstrationstudy, the concentration distributions were tailored so thatthe concentration of insulin increased monotonically fromone side of scaffold to the other, whereas b-GP phosphateconcentration decreased monotonically in the same direction.The use of both insulin and b-GP in conjunction with theirsystematically varied concentrations led to the differentiationof h-ADSCs in a location-dependent manner (higher chon-drocytic cell counts and increasing total collagen depositionwith increasing concentration of insulin) and different ex-tents of mineralization as generated by the b-GP concentra-tion distribution. The twin-screw extrusion electrospinningprocess, with its myriad capabilities for the tailoring ofspatial physical and chemical properties in the tissue engi-neering scaffolds, should serve as another enabler for themimicking of the gradations found in native tissues. Weanticipate that the potential of the new fabrication method ingenerating realistic functionally graded scaffolds will berecognized by other investigators who would utilize suchgraded scaffolds in vivo to demonstrate the true benefits offunctional grading.

Acknowledgments

We are grateful to Material Processing & Research Inc. ofHackensack, NJ, for making their MPR 7.5 mm twin-screwextrusion platform available to us. We thank David K.Moscatello from Coriell Institute for Medical Research(Camden, NJ) for providing the human adipose stromal cellline. We thank Dr. Halil Gevgilili of Highly Filled MaterialsInstitute for his contributions to the development of the hy-brid technology. We are also thankful to Markus F. Meyen-hofer for his help in histological analyses, to Dr. YeldaSertdemir Erisken for statistical evaluations, and to Prof.Xiaojun Yu of Stevens and Prof. Helen Lu of ColumbiaUniversity for their help, feedback, and comments.

Disclosure Statement

No competing financial interests exist.

References

1. Moffat, K.L., Sun, W.H., Pena, P.E., Chahine, N.O., Doty,S.B., Ateshian, G.A., Hung, C.T., and Lu, H.H. Character-ization of the structure-function relationship at the ligament-to-bone interface. Proc Natl Acad Sci USA 105, 7947, 2008.

2. McLauchlan, G.J., and Gardner, D.L. Sacral and iliac artic-ular cartilage thickness and cellularity: relationship to sub-chondral bone end-plate thickness and cancellous bonedensity. Rheumatology 41, 375, 2002.

3. Phillips, J.E., Burns, K.L., Le Doux, J.M., Guldberg, R.E., andGarcia, A.J. Engineering graded tissue interfaces. Proc NatlAcad Sci USA 105, 12170, 2008.

4. Bradley, D.A., Muthuvelu, P., Ellis, R.E., Green, E.M., At-tenburrow, D., Barrett, R., Arkill, K., Colridge, D.B., andWinlove, C.P. Characterisation of mineralisation of bone andcartilage: X-ray diffraction and Ca and SrK alpha X-rayfluorescence microscopy. Nucl Instrum Meth B 263, 1, 2007.

5. Jiang, J., Nicoll, S.B., and Lu, H.H. Co-culture of osteoblastsand chondrocytes modulates cellular differentiation in vitro.Biochem Biophys Res Commun 338, 762, 2005.

BIOMIMETIC SCAFFOLDS WITH GRADED MULTIPLE STIMULATORS 1249

6. Sarazin, P., Roy, X., and Favis, B.D. Controlled preparationand properties of porous poly(L-lactide) obtained from a co-continuous blend of two biodegradable polymers. Bioma-terials 25, 5965, 2004.

7. Ozkan, S., Kalyon, D.M., Yu, X., McKelvey C., andLowinger, M. Multifunctional polycaprolactone (PCL) scaf-folds integrated for controlled release and tissue engineer-ing: In vitro evaluation of released protein secondarystructure stability, release profile and biocompatibility. Bio-materials 30, 4336, 2009.

8. Luo, Y., Kobler, J.B., Zeitels, S.M., and Langer, R. Effects ofgrowth factors on extracellular matrix production by vocal foldfibroblasts in 3-dimensional culture. Tissue Eng 12, 3365, 2006.

9. Langer, R., and Vacanti, J.P. Tissue Engineering. Science 260,

920, 1993.10. Uebersax, L., Merkle, H.P., and Meinel, L. Insulin-like

growth factor I releasing silk fibroin scaffolds induce chon-drogenic differentiation of human mesenchymal stem cells.J Control Release 127, 12, 2008.

11. Li, C., Vepari, C., Jin, H.J., Kim, H.J., and Kaplan, D.L.Electrospun silk-BMP-2 scaffolds for bone tissue engineer-ing. Biomaterials 27, 3115, 2006.

12. Levenberg, S., Huang, N.F., Lavik, E., Rogers, A.B.,Itskovitz-Eldor, J., and Langer, R. Differentiation of humanembryonic stem cells on three-dimensional polymer scaf-folds. Proc Natl Acad Sci USA 100, 12741, 2003.

13. Wang, X., Wenk, E., Zhang, X., Meinel, L., Vunjak-Novakovic, G., and Kaplan, D.L. Growth factor gradientsvia microsphere delivery in biopolymer scaffolds for osteo-chondral tissue engineering. J Control Release 134, 81, 2009.

14. Benoit, D.S., Schwartz, M.P., Durney, A.R., and Anseth, K.S.Small functional groups for controlled differentiation ofhydrogel-encapsulated human mesenchymal stem cells. NatMater 7, 816, 2008.

15. Archer, R., and Williams, D.J. Why tissue engineering needsprocess engineering. Nat Biotechnol 23, 1353, 2005.

16. Erisken, C., Kalyon, D.M., and Wang, H.J. A hybridtwin screw extrusion/electrospinning method to processnanoparticle-incorporated electrospun nanofibres. Nano-technology 19, 165302, 2008.

17. Erisken, C., Kalyon, D.M., and Wang, H. Functionally gra-ded electrospun polycaprolactone and beta-tricalciumphosphate nanocomposites for tissue engineering applica-tions. Biomaterials 29, 4065, 2008.

18. Zuk, P.A., Zhu, M., Mizuno, H., Huang, J., Futrell, J.W.,Katz, A.J., Benhaim, P., Lorenz, H.P., and Hedrick, M.H.Multilineage cells from human adipose tissue: implicationsfor cell-based therapies. Tissue Eng 7, 211, 2001.

19. Hadhazy, C., and Dedukh, N.V. [Effect of insulin on cartilagedifferentiation in vitro]. Biull Eksp Biol Med 105, 219, 1988.

20. Julian, D., and Abbott, U.K. An avian model for comparativestudies of insulin teratogenicity. Anat Histol Embryol 27,

313, 1998.21. Marolt, D., Augst, A., Freed, L.E., Vepari, C., Fajardo, R.,

Patel, N., Gray, M., Farley, M., Kaplan, D., and Vunjak-Novakovic, G. Bone and cartilage tissue constructs grownusing human bone marrow stromal cells, silk scaffolds androtating bioreactors. Biomaterials 27, 6138, 2006.

22. Maor, G., Silbermann, M., von der, M.K., Heingard, D., andLaron, Z. Insulin enhances the growth of cartilage in organand tissue cultures of mouse neonatal mandibular condyle.Calcif Tissue Int 52, 291, 1993.

23. Allan, K.S., Pilliar, R.M., Wang, J., Grynpas, M.D., andKandel, R.A. Formation of biphasic constructs containing

cartilage with a calcified zone interface. Tissue Eng 13,

167, 2007.24. Chung, C.H., Golub, E.E., Forbes, E., Tokuoka, T., and

Shapiro, I.M. Mechanism of action of beta-glycerophosphateon bone cell mineralization. Calcif Tissue Int 51, 305, 1992.

25. Tenenbaum, H.C. Role of organic phosphate in mineraliza-tion of bone in vitro. J Dent Res 60 Spec No C, 1586, 1981.

26. Wall, M.E., Rachlin, A., Otey, C.A., and Loboa, E.G. Humanadipose-derived adult stem cells upregulate palladin duringosteogenesis and in response to cyclic tensile strain. Am JPhysiol Cell Physiol 293, C1532, 2007.

27. Wu, Y.N., Yang, Z., Hui, J.H., Ouyang, H.W., and Lee, E.H.Cartilaginous ECM component-modification of the micro-bead culture system for chondrogenic differentiation ofmesenchymal stem cells. Biomaterials 28, 4056, 2007.

28. Erisken, C., Kalyon, D.M., and Wang, H. Viscoelastic andbiomechanical properties of osteochondral tissue constructsgenerated from graded polycaprolactone and beta-tricalcium phosphate composites. J Biomech Eng 132,

091013, 2010.29. Lemke, M.J., Churchill, P.F., and Wetzel, R.G. Effect of

substrate and cell surface hydrophobicity on phosphateutilization in bacteria. Appl Environ Microbiol 61, 913, 1995.

30. Kondo, S., Xie, W.-F., Cha, S.H., and Sandell, L.J. Discoveryand characteristic features of cartilage-derived retinoic acid-sensitive protein (CD-RAP). Eur Cells Mater 8, 57, 1998.

31. Sakano, S., Zhu, Y., and Sandell, L.J. Cartilage-derived re-tinoic acid-sensitive protein and type II collagen expressionduring fracture healing are potential targets for Sox9 regu-lation. J Bone Miner Res 14, 1891, 1999.

32. Dietz, U.H., and Sandell, L.J. Cloning of a retinoic acid-sensitive mRNA expressed in cartilage and during chon-drogenesis. J Biol Chem 271, 3311, 1996.

33. Bosserhoff, A.K., and Buettner, R. Establishing the proteinMIA (melanoma inhibitory activity) as a marker for chon-drocyte differentiation. Biomaterials 24, 3229, 2003.

34. Dell’Accio, F., De Bari, C., and Luyten, F.P. Molecularmarkers predictive of the capacity of expanded human ar-ticular chondrocytes to form stable cartilage in vivo. ArthritisRheum 44, 1608, 2001.

35. Deng, C., Wynshaw-Boris, A., Zhou, F., Kuo, A., and Leder,P. Fibroblast growth factor receptor 3 is a negative regulatorof bone growth. Cell 84, 911, 1996.

36. Colvin, J.S., Bohne, B.A., Harding, G.W., McEwen, D.G., andOrnitz, D.M. Skeletal overgrowth and deafness in micelacking fibroblast growth factor receptor 3. Nat Genet 12,

390, 1996.37. Su, N., Yang, J., Xie, Y., Du, X., Lu, X., Yin, Z., Yin, L., Qi, H.,

Zhao, L., Feng, J., and Chen, L. Gain-of-function mutation ofFGFR3 results in impaired fracture healing due to inhibitionof chondrocyte differentiation. Biochem Biophys Res Com-mun 376, 454, 2008.

38. Zuscik, M.J., Hilton, M.J., Zhang, X., Chen, D., and O’Keefe,R.J. Regulation of chondrogenesis and chondrocyte differ-entiation by stress. J Clin Invest 118, 429, 2008.

39. Ghosh, P., and Smith, M. Osteoarthritis, genetic and mo-lecular mechanisms. Biogerontology 3, 85, 2002.

40. Lefebvre, V., Peeters-Joris, C., and Vaes, G. Modulation byinterleukin 1 and tumor necrosis factor alpha of productionof collagenase, tissue inhibitor of metalloproteinases andcollagen types in differentiated and dedifferentiated articu-lar chondrocytes. Biochim Biophys Acta 1052, 366, 1990.

41. Benya, P.D., Padilla, S.R., and Nimni, M.E. The progeny ofrabbit articular chondrocytes synthesize collagen types I and

1250 ERISKEN ET AL.

III and type I trimer, but not type II. Verifications by cyan-ogen bromide peptide analysis. Biochemistry 16, 865, 1977.

42. Yoon, Y.M., Kim, S.J., Oh, C.D., Ju, J.W., Song, W.K., Yoo,Y.J., Huh, T.L., and Chun, J.S. Maintenance of differentiatedphenotype of articular chondrocytes by protein kinase C andextracellular signal-regulated protein kinase. J Biol Chem277, 8412, 2002.

43. Bonaventure, J., Kadhom, N., Cohen-Solal, L., Ng, K.H.,Bourguignon, J., Lasselin, C., and Freisinger, P. Reexpressionof cartilage-specific genes by dedifferentiated human artic-ular chondrocytes cultured in alginate beads. Exp Cell Res212, 97, 1994.

44. Woodfield, T.B., van Blitterswijk, C.A., De, W.J., Sims, T.J.,Hollander, A.P., and Riesle, J. Polymer scaffolds fabricatedwith pore-size gradients as a model for studying the zonalorganization within tissue-engineered cartilage constructs.Tissue Eng 11, 1297, 2005.

45. Leong, K.F., Chua, C.K., Sudarmadji, N., and Yeong, W.Y.Engineering functionally graded tissue engineering scaf-folds. J Mech Behav Biomed Mater 1, 140, 2008.

46. Schaefer, D., Martin, I., Shastri, P., Padera, R.F., Langer, R.,Freed, L.E., and Vunjak-Novakovic, G. In vitro generation ofosteochondral composites. Biomaterials 21, 2599, 2000.

47. Sherwood, J.K., Riley, S.L., Palazzolo, R., Brown, S.C.,Monkhouse, D.C., Coates, M., Griffith, L.G., Landeen, L.K.,and Ratcliffe, A. A three-dimensional osteochondral com-posite scaffold for articular cartilage repair. Biomaterials 23,

4739, 2002.48. Ng, K.W., Ateshian, G.A., and Hung, C.T. Zonal Chon-

drocytes seeded in a layered agarose hydrogel create en-gineered cartilage with depth-dependent cellular andmechanical inhomogeneity. Tissue Eng Part A 15, 2315,2009.

49. Singh, M., Morris, C.P., Ellis, R.J., Detamore, M.S., andBerkland, C. Microsphere-based seamless scaffolds contain-ing macroscopic gradients of encapsulated factors for tissueengineering. Tissue Eng Part C Methods 14, 299, 2008.

50. Zeltinger, J., Sherwood, J.K., Graham, D.A., Mueller, R., andGriffith, L.G. Effect of pore size and void fraction on cellularadhesion, proliferation, and matrix deposition. Tissue Eng 7,

557, 2001.51. Li, X.R., Xie, J.W., Lipner, J., Yuan, X.Y., Thomopoulos, S.,

and Xia, Y.N. Nanofiber scaffolds with gradations in mineralcontent for mimicking the tendon-to-bone insertion site.Nano Lett 9, 2763, 2009.

52. Chang, Y.L., Stanford, C.M., and Keller, J.C. Calcium andphosphate supplementation promotes bone cell mineraliza-tion: implications for hydroxyapatite (HA)-enhanced boneformation. J Biomed Mater Res 52, 270, 2000.

53. Yu, H., and Grainger, D.W. Modified release of hydrophilic,hydrophobic and peptide agents from ionized amphiphilicgel networks. J Control Release 34, 117, 1995.

54. Serravezza, J.C., Endo, F., Dehaan, R.L., and Elsas, L.J.Insulin-induced receptor loss reduces responsiveness ofchick heart-cells to insulin. Endocrinology 113, 497, 1983.

55. Bhattarai, N., Edmondson, D., Veiseh, O., Matsen, F.A., andZhang, M.Q. Electrospun chitosan-based nanofibers andtheir cellular compatibility. Biomaterials 26, 6176, 2005.

56. Kemp, S.F., Mutchnick, M., and Hintz, R.L. Hormonal-control of protein-synthesis in chick chondrocytes—acomparison of effects of insulin, somatomedin-c and triio-dothyronine. Acta Endocrinol 107, 179, 1984.

57. Muramatsu, T., Kita, K., Nakagawa, S., and Okumura, J.Effects of insulin, growth-hormone, IGF-I and IGF-II on

phenylalanine extraction in the chicken-embryo cul-tured in vitro. Comp Biochem Physiol A Physiol 104,

507, 1993.58. Scott, J.E., and Dorling, J. Differential staining of acid gly-

cosaminoglycans (mucopolysaccharides) by alcian blue insalt solutions. Histochemie 5, 221, 1965.

59. Zhu, W., Mow, V.C., Koob, T.J., and Eyre, D.R. Viscoelasticshear properties of articular cartilage and the effects of gly-cosidase treatments. J Orthop Res 11, 771, 1993.

60. Steinert, A., Weber, M., Dimmler, A., Julius, C., Schutze, N.,Noth, U., Cramer, H., Eulert, J., Zimmermann, U., andHendrich, C. Chondrogenic differentiation of mesenchymalprogenitor cells encapsulated in ultrahigh-viscosity alginate.J Orthop Res 21, 1090, 2003.

61. Neumann, K., Dehne, T., Endres, M., Erggelet, C., Kaps, C.,Ringe, J., and Sittinger, M. Chondrogenic differentiationcapacity of human mesenchymal progenitor cells derivedfrom subchondral cortico-spongious bone. J Orthop Res 26,

1449, 2008.62. Junker, J.P.E., Sommar, P., Skog, M., Johnson, H., and Kratz,

G. Adipogenic, chondrogenic and osteogenic differentiationof clonally derived human dermal fibroblasts. Cells TissuesOrgans 191, 105, 2010.

63. Williams, R., Khan, I.M., Richardson, K., Nelson, L.,McCarthy, H.E., Analbelsi, T., Singhrao, S.K., Dowthwaite,G.P., Jones, R.E., Baird, D.M., Lewis, H., Roberts, S., Shaw,H.M., Dudhia, J., Fairclough, J., Briggs, T., and Archer, C.W.Identification and clonal characterisation of a progenitor cellsub-population in normal human articular cartilage. PLoSONE 5, e13246, 2010.

64. Garreta, E., Genove, E., Borros, S., and Semino, C.E. Osteo-genic differentiation of mouse embryonic stem cells andmouse embryonic fibroblasts in a three-dimensional self-assembling peptide scaffold. Tissue Eng 12, 2215, 2006.

65. Knabe, C., Kraska, B., Koch, C., Gross, U., Zreiqat, H., andStiller, M. A method for immunohistochemical detection ofosteogenic markers in undecalcified bone sections. BiotechHistochem 81, 31, 2006.

66. Schutze, N., Noth, U., Schneidereit, J., Hendrich, C., andJakob, F. Differential expression of CCN-family members inprimary human bone marrow-derived mesenchymal stemcells during osteogenic, chondrogenic and adipogenic dif-ferentiation. Cell Commun Signal 3, 5, 2005.

67. Sophia Fox, A.J., Bedi, A., and Rodeo, S.A. The basic scienceof articular cartilage: structure, composition, and function.Sports Health: A Multidisciplinary Approach 1, 461, 2009.

68. Barsh, G.S., David, K.E., and Byers, P.H. Type I osteogenesisimperfecta: a nonfunctional allele for pro alpha 1 (I)chains of type I procollagen. Proc Natl Acad Sci USA 79,

3838, 1982.69. Kirsch, E., Krieg, T., Nerlich, A., Remberger, K., Meinecke,

P., Kunze, D., and Muller, P.K. Compositional analysis ofcollagen from patients with diverse forms of osteogenesisimperfecta. Calcif Tissue Int 41, 11, 1987.

70. Junqueira, L.C., Bignolas, G., and Brentani, R.R. Picrosiriusstaining plus polarization microscopy, a specific method forcollagen detection in tissue sections. Histochem J 11, 447,1979.

71. Zuk, P.A., Zhu, M., Ashjian, P., De Ugarte, D.A., Huang, J.I.,Mizuno, H., Alfonso, Z.C., Fraser, J.K., Benhaim, P., andHedrick, M.H. Human adipose tissue is a source of multi-potent stem cells. Mol Biol Cell 13, 4279, 2002.

72. Awad, H.A., Wickham, M.Q., Leddy, H.A., Gimble,J.M., and Guilak, F. Chondrogenic differentiation of

BIOMIMETIC SCAFFOLDS WITH GRADED MULTIPLE STIMULATORS 1251

adipose-derived adult stem cells in agarose, alginate, andgelatin scaffolds. Biomaterials 25, 3211, 2004.

73. Tapp, H., Deepe, R., Ingram, J.A., Kuremsky, M., Hanley,E.N., Jr., and Gruber, H.E. Adipose-derived mesenchymalstem cells from the sand rat: transforming growth factor betaand 3D co-culture with human disc cells stimulate proteo-glycan and collagen type I rich extracellular matrix. ArthritisRes Ther 10, R89, 2008.

74. Musselmann, K., Kane, B., Alexandrou, B., and Hassell, J.R.Stimulation of collagen synthesis by insulin and proteogly-can accumulation by ascorbate in bovine keratocytes in vitro.Invest Ophthalmol Vis Sci 47, 5260, 2006.

75. Bosserhoff, A.K., Kondo, S., Moser, M., Dietz, U.H., Cope-land, N.G., Gilbert, D.J., Jenkins, N.A., Buettner, R., andSandell, L.J. Mouse CD-RAP/MIA gene: structure, chro-mosomal localization, and expression in cartilage andchondrosarcoma. Dev Dyn 208, 516, 1997.

76. Guba, M., Bosserhoff, A.K., Steinbauer, M., Abels, C., An-thuber, M., Buettner, R., and Jauch, K.W. Overexpression ofmelanoma inhibitory activity (MIA) enhances extravasationand metastasis of A-mel 3 melanoma cells in vivo. Br JCancer 83, 1216, 2000.

77. Moser, M., Bosserhoff, A.K., Hunziker, E.B., Sandell, L.,Fassler, R., and Buettner, R. Ultrastructural cartilage abnor-malities in MIA/CD-RAP-deficient mice. Mol Cell Biol 22,

1438, 2002.78. Valverde-Franco, G., Binette, J.S., Li, W., Wang, H., Chai, S.,

Laflamme, F., Tran-Khanh, N., Quenneville, E., Meijers, T.,Poole, A.R., Mort, J.S., Buschmann, M.D., and Henderson,J.E. Defects in articular cartilage metabolism and early ar-thritis in fibroblast growth factor receptor 3 deficient mice.Hum Mol Genet 15, 1783, 2006.

79. Valcourt, U., Gouttenoire, J., Moustakas, A., Herbage, D.,and Mallein-Gerin, F. Functions of transforming growthfactor-beta family type I receptors and smad proteins in thehypertrophic maturation and osteoblastic differentiation ofchondrocytes. J Biol Chem 277, 33545, 2002.

80. Tenenbaum, H.C. Levamisole and inorganic pyrophosphateinhibit beta-glycerophosphate induced mineralization ofbone formed in vitro. Bone Miner 3, 13, 1987.

81. Gough, J.E., Jones, J.R., and Hench, L.L. Nodule formationand mineralisation of human primary osteoblasts cultured ona porous bioactive glass scaffold. Biomaterials 25, 2039, 2004.

82. Fujita, T., Meguro, T., Izumo, N., Yasutomi, C., Fukuyama,R., Nakamuta, H., and Koida, M. Phosphate stimulates dif-ferentiation and mineralization of the chondroprogenitorclone ATDC5. Jpn J Pharmacol 85, 278, 2001.

83. Kandel, R.A., Boyle, J., Gibson, G., Cruz, T., and Speagle, M.In vitro formation of mineralized cartilagenous tissue byarticular chondrocytes. In Vitro Cell Dev Biol Anim 33, 174,1997.

84. Wang, C.C., Hung, C.T., and Mow, V.C. An analysis of theeffects of depth-dependent aggregate modulus on articularcartilage stress-relaxation behavior in compression. J Bio-mech 34, 75, 2001.

85. Armstrong, C.G., Lai, W.M., and Mow, V.C. An analysis ofthe unconfined compression of articular cartilage. J BiomechEng 106, 165, 1984.

86. Park, S., Hung, C.T., and Ateshian, G.A. Mechanical re-sponse of bovine articular cartilage under dynamic uncon-fined compression loading at physiological stress levels.Osteoarthritis Cartilage 12, 65, 2004.

87. Garnero, P., Borel, O., Gineyts, E., Duboeuf, F., Solberg, H.,Bouxsein, M.L., Christiansen, C., and Delmas, P.D. Extra-cellular post-translational modifications of collagen are ma-jor determinants of biomechanical properties of fetal bovinecortical bone. Bone 38, 300, 2006.

88. Burr, D.B. The contribution of the organic matrix to bone’smaterial properties. Bone 31, 8, 2002.

Address correspondence to:Dilhan M. Kalyon, Ph.D.

Department of Chemical Engineering and Material Scienceand Department of Chemistry, Chemical Biology

and Biomedical EngineeringStevens Institute of Technology

McLean Chemical and Life Sciences Building1 Castle Point on Hudson

Hoboken, NJ 07030

E-mail: dkalyon@stevens.edu

Received: October 21, 2009Accepted: December 23, 2010

Online Publication Date: January 31, 2011

1252 ERISKEN ET AL.